Contribution to chromosome numbers and phylogeny of Turkish Vincetoxicum Wolf (Apocynaceae, Asclepiadoideae)

We report chromosome counts for ten taxa of Vincetoxicum sensu stricto (s. str.) (Apocynaceae) from Turkey (of which two are endemic), including the fi rst chromosome counts for V. canescens subsp. pedunculata, V. funebre, V. fuscatum subsp. boissieri, V. parvifl orum and V. tmoleum. Two taxa of V. fuscatum proved to be tetraploid (2n=44) and the remaining eight taxa diploid (2n=22). Molecular phylogenetic analyses based on nrDNA (ITS) and cpDNA (trnT-trnL) (including 31 newly generated sequences) confi rm the position of the Turkish Vincetoxicum in the Vincetoxicum s. str. clade. Vincetoxicum fuscatum, V. parvifl orum, V. speciosum, as well as the Turkish endemic V. fuscatum subsp. boissieri, were clearly resolved as species-level clades, whereas the delimitation of the rest of the Turkish taxa was less clear based on molecular data.


Introduction
Vincetoxicum Wolf is one of the largest and most widespread genera of the subfamily Asclepiadoideae (Apocynaceae). Molecular data indicated that Vincetoxicum sensu lato (s. lat.), comprising ca 150 species, is widely distributed from Australasia and the Far East to Africa, Europe, and North America (Liede-Schumann et al. 2016). The well-defined Vincetoxicum s. str. clade with ca 45 species extends from the Far East via the Asian mountain ranges into Central Europe and was also introduced in North America (Liede-Schumann et al. 2016). Vincetoxicum displays the flowers typical of the subfamily with gynostegium and five pollinaria, its sterile corona is staminal and characteristically composed of five fleshy lobes separated near the base or partly connected by smaller interstaminal parts (Liede 1996). In addition, most Vincetoxicum individuals have been determined to be poisonous to humans, and some have been used in conventional and folk medicine as well as in modern medicine (Zaidi & Crow 2005;Mansoor et al. 2011).
Chromosome number data remain valuable today in systematics, and establishing which plants are polyploids and which are diploids is important to allow for a better understanding of plants to be included in phylogenetic studies. Polyploidy, widely accepted as a mechanism of reproductive isolation and plant speciation, has been an important force shaping the evolutionary history of vascular plants. It is estimated that between 15% and 30% of speciation in angiosperms results from polyploidy. Over the last decade, there has been an increase in studies investigating how polyploids influence phylogenetic community structure by combining ploidal information with phylogenetic analyses of plant communities. The polyploid and diploid species have been identified as having phenotypic differences and molecular variations associated with whole genome duplication. While these differences are considered to increase the ecological success of polyploids in novel communities, it remains relatively unexplored whether polyploids are indeed better competitors in a community context (Gregg et al. 2017;Gaynor et al. 2018). For nearly a century, botanists in particular, have been interested in the determination and documentation of chromosome numbers which have been extensively utilized as an important phylogenetic character in the context of cytotaxonomy (Rice et al. 2015). Although these data have been documented along the years, to date, chromosome number data of only approximately 25% of flowering plants have been determined (Stace 2000;Garbari et al. 2012). Despite the great floristic richness of Turkey, only 15% of the vascular plant taxa have had their chromosome number investigated (Vladimirov et al. 2015).
The aim of the present study was: 1) to report the chromosome number of Turkish representatives of Vincetoxicum s. str. since very few counts have been reported in the literature previously; and 2) to contribute to the phylogenetic position of Vincetoxicum s. str. (of which two are endemic) based on newly generated sequences from Turkish accessions.

Sampling
Locality information of samples used for chromosome counting and molecular studies is presented in Appendix 1. Of these, 16 representatives belonging to 10 Turkish taxa of Vincetoxicum s. str. were used in somatic chromosomal studies. In total, 59 accessions (Vincetoxicum s. str. (53), Cionura erecta L. (3), Cynanchum acutum L. (2) and Gomphocarpus fruticosus (L.) W.T.Aiton (1)) were used for molecular analysis (Appendix 1). Of these, one Cynanchum, three Cionura and 31 Vincetoxicum s. str. members were collected by the present authors from their natural habitats in Turkey (Appendix 1). All specimens were first processed using the standard herbarium techniques given by Woodland (1997); they were identified using the Flora of Turkey (Browicz 1978), Flora of Russia (Pobedimova 1952), Flora Europaea (Markgraf 1972), and stored at the Herbarium of Recep Tayyip Erdogan University, Department of Biology (RUB).

Cytological analyses
For somatic chromosomal examination, the mature seeds were germinated on wet filter paper in Petri dishes at +27°C. Active roots were cut at 1-1.5 cm from the tips and these were pretreated for 16 hours in α-monobromonaphthalene at +4°C and then fixed using Carnoy solution (3:1 absolute alcohol:glacial acetic acid) overnight. Fixed root tips were transferred to 70% alcohol and stored at +4°C until analysis. Afterwards, the root tips were hydrolyzed with 1 N HCl for 12 minutes at 60°C and stained with 2% aceto orcein for 24 hours at room temperature. Stained root tips were squashed in a drop of 45% acetic acid and the preparations were mounted in entellan in order to obtain permanent slides. The best metaphase plates, including at least ten well-spread cells, were photographed with an Olympus BX51 microscope with an attached digital camera, and they were also drawn from permanent slides. For chromosome counting, both the 10 × 100 enlarged photographs and the drawings were used (Jones & Rickards 1991;Elçi 1994;Martin et al. 2019).

DNA isolation, PCR amplification, and sequencing
Total genomic DNA was extracted from silica dried leaves following the modified CTAB extraction procedure of Doyle & Doyle (1987). The nrITS region and trnT-trnL spacer were amplified using the universal 'ITS4 and ITS5' primers (White et al. 1990) and the universal 'a and b' primers (Taberlet et al. 1991), respectively, according to the PCR conditions described by Gültepe et al. (2010). The PCR products were sequenced with the aid of Macrogen Inc. (Seoul, Korea) using the same primers.

DNA sequence alignment and phylogenetic analyses
The nucleotide sequences were aligned by ClustalW using BioEdit v.7.0 software (Hall 1999). The obtained sequence data were compared with GenBank sequences using BLAST and manually verified. For both ITS and trnT-trnL sequences, total nucleotide length (bp), parsimony informative sites and the interspecific and intraspecific pairwise differences from a distance matrix were calculated using MEGA v.7.0 (Kumar et al. 2016).
The phylogenetic analysis included 52 Vincetoxicum taxa attributed to the 'Vincetoxicum s. str. clade', as well as one accession attributed to the 'Far Eastern clade' according to Liede-Schumann et al. (2016). Of these, 31 accessions belonged to 10 taxa of Turkish Vincetoxicum s. str. (Appendix 1), and the dataset of ITS and trnT-trnL belonging to the remaining 22 accessions studied by Liede-Schumann et al. (2016) was retrieved from GenBank (www.ncbi.nlm.nih.gov). Our sampling aimed to cover the area of the genus Vincetoxicum s. str. in Turkey as completely as possible, with particular emphasis on the 'Vincetoxicum s. str. clade' including the taxa attributed to the European, Western Irano-Turanian, Eastern Mediterranean, Southern Russian, Eastern Irano-Turanian, and Western Himalayan clades (Liede-Schumann et al. 2016), which are neighbouring regions to Turkey. As outgroup, six specimens belonging to three different species of Asclepiadoideae (Cionura erecta, Cynanchum acutum and Gomphocarpus fruticosus), which are also distributed in Turkey, were selected for the phylogenetic analysis (Appendix 1).
The two dataset (ITS and trnT-trnL) were combined according to the ILD test results of Liede-Schumann et al. (2012). Phylogenetic relationships were reconstructed using Bayesian Inference (BI) analyses and Maximum Parsimony (MP). For Bayesian Inference (BI) analyses, evolutionary models were assessed with Akaike information criterion (AIC) as implemented in MrModeltest 2.3 software (Nylander 2004). The most suitable models, as detected by AIC, were GTR+G for the nrDNA ITS marker and GTR+I for the cpDNA trnT-trnL marker. The BI methods were carried out using MrBayes ver. 3.1 (Ronquist & Huelsenbeck 2003). For the BI analyses, a Metropolis-coupled Markov chain Monte Carlo (MCMCMC) algorithm was employed with two simultaneous runs of four parallel MCMCMC each for 1 million generations, starting with a random tree. The trees were saved every 1000 th generation, and the first 20% were discarded as 'burn-in'. The remaining trees were used to estimate the majority rule consensus tree and Bayesian posterior probabilities (PP).
Maximum Parsimony (MP) methods were implemented in the PAUP* ver. 4.0b10 (Swofford 2003) software. For the MP analyses, a script was first created using the parsimony ratchet (Nixon 1999) method implemented in the PRAP v.2.0 (Müller 2004). The script contains standard ratchet settings (200 ratchet iterations with 25% of the positions randomly upweighted (weight = 2) during each replicate and 10 random additional cycles). This file, containing the sequence data and commands, was analysed by Jackknife (JK) in PAUP* ver. 4.0b10 (Swofford 2003) to calculate JK support values for the branches. Then, MP analysis in PAUP* ver. 4.0b10 (Swofford 2003) was performed with 1000 repetitions according to the majority rule. The JK support values were transferred to the phylogenetic tree obtained from the results of the MP analysis. Indels were coded as informative characters according to the Simple Indel Coding (SIC) method (Simmons & Ochoterena 2000) as implemented in the programme SeqState ver. 1.40 (Müller 2005) and added at the end of the sequence dataset.
In phylogenetic analyses based on sequence data, monophyly, branch lengths, branch support and genealogical concordance have been reported as often-used criteria for species delimitation (Leliaert et al. 2014). Our phylogeny based on nrDNA (ITS) and cpDNA (trnT-trnL) sequences contains multiple specimens per species, and species are delimited based on topological criteria, monophyly and branch support. Additionally, some typical floral characteristics of Vincetoxicum s. str., which were not analysed, were also presented in our phylogenetic tree in order to see whether there is a concordance between molecular data and these diagnostic characteristics.

Chromosome numbers
Chromosome counts for 16 accessions belonging to ten taxa of Vincetoxicum s. str. from Turkey are presented in Figs 1-2. The somatic chromosome numbers were determined as 2n=4x=44 in two subspecies   of V. fuscatum (subsp. fuscatum and subsp. boissieri) and 2n=2x=22 in the rest of the examined taxa. To the best of our knowledge, these are the first chromosome counts for V. canescens subsp. pedunculata, V. funebre, V. fuscatum subsp. boissieri, V. parviflorum and V. tmolem (Table 1).

Phylogenetic analyses
The topologies of the phylogenetic trees inferred from the concatenated alignment of nrDNA ITS + trnT-trnL using MP and BI analyses were similar. Therefore, the Bayesian majority rule tree is shown in Fig.  3, along with JK support values of the MP analysis. The phylogenetic tree revealed that all the examined taxa of Vincetoxicum were separated from the outgroup (which included specimens belonging to the three separate species: Cionura erecta, Cynanchum acutum and Gomphocarpus fruticosus) and formed a clade composed of two subclades, I and II. Clade I comprised only V. atratum corresponding to the 'Far Eastern clade' in Liede-Schumann et al. (2016), and clade II, corresponding to the 'Vincetoxicum s. str. clade' in Liede-Schumann et al. (2016), included all the remaining ingroup taxa. Our phylogenetic tree (Fig. 3) revealed that among the examined taxa of Turkish Vincetoxicum s. str., only V. fuscatum, Fig. 3. Phylogenetic tree of Vincetoxicum s. str., including Turkish samples (in bold) based on the combined dataset (ITS and trnT-trnL). The support values on branches indicate Jackknife (JK) and Posterior Probability (PP) higher than 50% and 0.7, respectively. Clade designations correspond to Liede-Schumann et al. (2016). Refer to Appendix 1 for accession abbreviations (G:Güven, M:Makbul).
V. parviflorum and V. speciosum were recovered as monophyletic, whereas the delimitation of the remaining six taxa was less clear based on molecular data. In the phylogenetic tree, some typical floral characteristics of the investigated taxa of Vincetoxicum s. str. such as the indumentum of the corolla and shape of the corona (which were not analysed) were also presented. Nevertheless, these characters seemed to be homoplastic, since they appeared in different subclades.

Discussion
In the present study, the chromosome number for ten taxa of Vincetoxicum s. str. from Turkey, including two endemics, are reported. In addition, the phylogenetic positions of V. fuscatum subsp. boissieri and V. parviflorum, which are endemic to Turkey, were investigated based on nrDNA ITS and cpDNA trnT-trnL with 31 newly generated sequences.
In the Flora of Turkey (Browicz 1975), the yellow flowered V. canescens was represented by two subspecies, subsp. canescens and subsp. pedunculata, which were shown in the present study to be diploid according to two Turkish accessions. The chromosome number of the other yellowish flowered Turkish Vincetoxicum s. str., V. tmoleum, was determined as 2n=2x=22 from two Turkish accessions. To the best of our knowledge, these are the first chromosome counts for V. canescens subsp. pedunculata and V. tmoleum. Similarly, V. canescens subsp. canescens has previously been reported as diploid (2n=2x=22) in the Flora of China (Li et al. 1995). Our chromosomal data on V. canescens subsp. canescens originating from Turkey is in accordance with these previous results.
According to our phylogenetic analyses, V. canescens and V. tmoleum also clustered in the same group (including V. creticum) corresponding to the 'Eastern Mediterranean clade' in Liede-Schumann et al. (2016). It was reported that these two species had many morphological and palynological characteristics in common, such as densely hairy corolla and deeply parted corona segments (Güven 2017), an ovate pollinium and corpusculum shape, and a rugulate pollen surface (Güven et al. 2015). The similarity in floral morphology between V. canescens and V. tmoleum was also reported by Browicz (1975). In addition, the same chromosome number was determined for V. canescens and V. tmoleum. However, the stem and fruit morphology of the two species are quite different. Vincetoxicum canescens, characterised by decumbent stems with a canescent-tomentose indumentum and ovoid fruits, is easily distinguished from V. tmoleum, which has erect and crisped hairy stems and slender ovoid fruits (Browicz 1975;Güven 2017). Vincetoxicum creticum (endemic to Crete-Greece) clustered with V. tmoleum in the same sub-clade. This species was previously reported as morphologically close to V. tmoleum by Browicz (1975). Our molecular results support this view.
Two subspecies of V. canescens, subsp. canescens and subsp. pedunculata, differing from each other by the length of their peduncle (Browicz 1978), are closely related taxa in some anatomical and palynological aspects (Güven 2017). Our analyses showed that these two subspecies had identical chromosome numbers of 2n=22 (diploid), and had identical sequences. Further molecular studies involving more gene regions will be needed to resolve the identity of the two subspecies.
Vincetoxicum fuscatum was represented by two morphologically close subspecies, subsp. boissieri and subsp. fuscatum, distinguished only by the corolla indumentum in the Flora of Turkey (Browicz 1975). Our analyses show that the two subspecies are tetraploid. While this was the first chromosome count for the subsp. boissieri, the chromosome number of 2n=22 (diploid) has been reported for the subsp. fuscatum from Greece (Strid & Franzen 1981). The intraspecific differentiation of the ploidy level within the plant species might be a result of the geographical distribution of the taxa (Morawetz 1984). In support of this, it has been reported that plant taxa collected from various geographical regions can have different chromosome numbers (Ozcan et al. 2008). Similarly, the chromosome number of V. nigrum from Spain has been reported as 2n=2x=22 (diploid) (Aparicio & Silvestre 1985), whereas it was counted as 2n=4x=44 (tetraploid) in populations distributed in both the Netherlands (Van den Brand et al. 1979) and Canada (Liede-Schumann et al. 2012). These differences can be caused by the variations in populations of V. nigrum and in environmental conditions between different geographical regions.
The taxa of V. fuscatum clustered under the same subclade together with V. maeoticum from Russia and V. intermedium from Kazakhstan (Fig. 3), corresponding to the 'Southern Russian clade' of Liede-Schumann et al. (2016). Marhold (2011) reported that V. maeoticum and V. intermedium are probably synonyms of V. fuscatum subsp. fuscatum. This view is supported by our molecular results. Browicz (1978) recognised that V. fuscatum subsp. boissieri and V. fuscatum subsp. fuscatum, which were distinguished only by their corolla indumentum, were separated at the subspecies level based on high morphological similarities. However, Pobedimova (1952) previously treated these two taxa as separate species named Antitoxicum boissieri (Kusn.) Pobed. (= V. fuscatum subsp. boissieri) and Antitoxicum minus (K.Koch) Pobed. (= V. fuscatum subsp. fuscatum). Our karyological analyses showed that these two subspecies of V. fuscatum had identical chromosome numbers of 2n=44 (tetraploid). Otherwise, in our molecular analyses, three specimens of subsp. boissieri formed a well-supported subclade among the unresolved specimens of V. fuscatum subsp. fuscatum, V. maeoticum and V. intermedium. We therefore suggest to raise V. fuscatum subsp. boissieri to the species level sister to V. fuscatum according to the present molecular data.
The somatic chromosome number of V. hirundinaria subsp. hirundinaria, which was the first report from Turkish accessions, was determined as 2n=2x=22. Our results are consistent with the previous counts reported by Uhríková et al. (1985) in Slovakia and Liede-Schumann et al. (2012) in Germany. The present study revealed a diploid chromosome number of 2n=2x=22 for V. scandens and V. speciosum from Turkish accessions. These findings are in agreement with the previous results of V. scandens in Iran (Lessani & Chariat-Panahi 1979) and V. speciosum in Italy (Pardi 1933). In this study, the chromosome number of V. funebre was determined for the first time as 2n=2x=22 from a Turkish accession.
Vincetoxicum hirundinaria subsp. hirundinaria, characterised by white-flowers, clustered with V. jailicola, V. rossicum and V. schmalhausenii (the European clade, according to Liede-Schumann et al. 2016), forming a clade in sister-group position to V. funebre, V. rehmannii and V. scandens (the Irano-Turanian clade according to Liede-Schumann et al. 2016), as well as to V. speciosum. The present molecular results are in accordance with the chromosome counts as well as some morphological and palynological features of individuals of the four Turkish Vincetoxicum taxa (V. funebre, V. hirundinaria subsp. hirundinaria, V. scandens and V. speciosum). Güven (2017) also reported that these taxa were characterised by a densely or sparsely hairy corolla with cup-shaped corona and obovate pollinia.
Vincetoxicum scandens, characterised by twining stems up to 2 m high, clustered with V. funebre, which is known to have erect and shorter (40-135 cm) stems. However, these taxa have some floral features in common, such as hairy corolla and cup-shaped five-parted corona (Pobedimova 1952;Güven 2017). Furthermore, V. funebre and V. scandens have similar palynological (obovate pollinium and oblong corpusculum) characteristics (Güven et al. 2015) and the same chromosome number of 2n=2x=22.
Vincetoxicum speciosum accessions formed a distinct subclade. In Turkey, V. hirundinaria subsp. hirundinaria and V. speciosum share the same overlapping humid habitats such as Quercus-forest openings in northwest Anatolia and Thrace regions. However, these two diploid representatives of Vincetoxicum can easily be differentiated by their morphological and palynological characteristics. While V. speciosum is characterised by a blackish purple corolla and an erect stem with wholly velutinous pubescence, V. hirundinaria subsp. hirundinaria has a white corolla and crisped hairy stems that occasionally twine near the apex (Browicz 1975). Güven et al. (2015) noted that the pollinia surface of V. speciosum exhibited gemmate ornamentation, in contrast to the rugulate surface in V. hirundinaria subsp. hirundinaria. Furthermore, V. speciosum was characterised by the longest obovate pollinium among the Turkish Vincetoxicum s. str. taxa (Güven et al. 2015).
Vincetoxicum nigrum is characterised by a weak twining stem and dark blackish-purple coloured and hairy flowers. Browicz (1978) indicated that the records previously reported from Anatolia for V. nigrum were incorrect and so they were treated under V. scandens. Vincetoxicum nigrum is morphologically similar to V. scandens but differs from this species by its weak growth, smaller leaves, shorter peduncles (not more than 1 cm), and cup-shaped corona with small coronal teeth (Markgraf 1972;Browicz 1978). Additionally, while both di-(2n=22) and tetraploid (2n=44) chromosome numbers have been counted for V. nigrum (Liede-Schumann et al. 2012), the chromosome counts for V. scandens, reported in both the present and previous (Lessani & Chariat-Panahi 1979) studies, were all 2n=22 (diploid). These two species are also located in different subclades according to the present molecular results.
In the present study, the chromosome number of V. parviflorum was determined for the first time as 2n=2x=22 from a Turkish accession. In the Flora of Turkey, Browicz (1978) reported that V. parviflorum could be a small-flowered variety of V. fuscatum due to their similar flowers and habits. However, Boissier (1875), who treated V. fuscatum and V. parviflorum as different at the species level, noted that these two taxa differ from each other in a number of morphological characters related to the stem and the flower. In accordance with Boissier (1875), Güven et al. (2015) also determined that although V. parviflorum exhibited an elliptic pollinium, two subspecies of V. fuscatum were characterised by a clavate pollinium. While the somatic chromosome number was determined as 2n=22 (diploid) in V. parviflorum, it was counted as 2n=44 (tetraploid) for both subspecies of V. fuscatum. Our molecular analysis also showed that V. parviflorum separated from taxa of V. fuscatum and clustered in a different subclade with V. assadii, V. mozaffarianii and V. pumilum (Eastern Irano-Turanian taxa according to Liede-Schumann et al. (2016)).
In conclusion, the results of our phylogenetic analyses, including 31 newly generated ITS and trnT-trnL sequences belonging to 10 taxa of Vincetoxicum s. str. from Turkey, agrees largely with previous phylogenetic results (Liede-Schumann et al. 2016). In addition to the previous study, DNA sequences were determined for V. fuscatum subsp. boissieri and V. parviflorum endemic to Turkey, as well as V. speciosum of European origin. Of the studied taxa, V. fuscatum, V. parviflorum and V. speciosum are clearly resolved at the species level, whereas the delimitation of the rest of the Turkish taxa was less clear based on molecular data. Three specimens of the endemic V. fuscatum subsp. boissieri formed a wellsupported subclade within the clade including unresolved specimens of V. fuscatum subsp. fuscatum, V. maeoticum and V. intermedium. We therefore suggest that V. fuscatum subsp. boissieri, previously treated as Antitoxicum boissieri by Pobedimova (1952), should be raised to the species level, sister to V. fuscatum. The other endemic taxon V. parviflorum, reported as a morphologically related taxon to V. fuscatum by Browicz (1978), separated from the V. fuscatum taxa and clustered in a different clade, sister to V. assadii, V. mozaffarianii and V. pumilum from Iran. The molecular results were evaluated together with some typical floral characteristics of Vincetoxicum s. str., such as the indumentum of the corolla and shape of the corona (which were not analysed). Nevertheless, these characters seemed to be homoplastic, since they appeared in different subclades.